Recently, CMOS image sensors have been widely used in digital still cameras, camcorders, surveillance cameras, and the like, and the market for the CMOS image sensors has been expanded. Each pixel in a CMOS image sensor converts incident light into electrons by a photodiode as a photoelectric conversion device, and accumulates the electrons for a certain period, and then outputs a signal corresponding to the amount of accumulated charge to an analog-digital (AD) converter contained in a chip. The AD converter digitalizes the signal, and then outputs the digitalized signal to a stage following the AD converter. In the CMOS image sensor, for image pickup, such pixels are arranged in a matrix form.
FIG. 1 is a diagram illustrating a typical chip configuration of a CMOS image sensor that is a solid-state image pickup device.
This CMOS image sensor 10 includes a pixel array section 11, a row drive circuit 12, AD converters 13, switches 14, an output circuit 15, row control lines 16, vertical signal lines 17, and a transfer line 18.
In the pixel array section 11, a plurality of pixels PX are arranged along a row direction and a column direction in a matrix form, and the vertical signal line 17 is shared by a plurality of pixels PX arranged along the column direction, and is connected to the AD converter 13 arranged corresponding to each column. On the other hand, the row drive circuit 12 selects only one row from a plurality of rows, and drives the row control line 16 to read the accumulated charges from the pixels PX in the selected row. The row control line 16 is configured of one or a plurality of control lines to execute such reading from the pixels or resetting of the pixels. As used herein, the term “resetting” refers to an operation in which the accumulated charges are discharged from the pixels to return the pixels to a state before exposure, and, for example, the resetting may be executed as a shutter operation immediately after reading from each row or when exposure starts. At the time of reading, analog signals transmitted to the AD converter 13 through the vertical signal line 17 are converted into digital signals, and the digital signals are sequentially transmitted to the output circuit 15 through the switch 14 to be output to an image processing unit located inside or outside the chip that is not illustrated.
When the CMOS image sensor 10 completes reading from one row in such a manner, a next row is selected, and reading, AD conversion, and outputting are repeated in a similar manner. When processing on all of the rows is completed, outputting of one frame of image data is completed.
On the other hand, PTL 1 has proposed a novel technique of counting photons in a time-divisional manner. In the counting technique, a binary decision as to whether or not a photon is incident on a photodiode in a certain period is repeatedly performed a plurality of times, and results of the binary decisions are integrated to obtain two-dimensional image pickup data. In other words, a signal from the photodiode in each certain period (each unit exposure period) is sensed, and when one or more photons are incident on the photodiode in the period, a counter connected to each pixel counts up by 1 irrespective of the number of incident photons. If the frequency of photon incidence is random along a time axis, the actual number of incident photons and the number of counts follow the Poisson distribution; therefore, when the frequency of incidence is low, the actual number of incident photons and the number of counts have a substantially linear relationship, and when the frequency of incidence is high, an output is compressed.
Moreover, PTL 2 has proposed a technique of improving an aperture ratio of pixels by separating a sense circuit and a counter circuit for the above-described time-divisional photon counting from the pixels and hierarchically arranging them.
An image sensor using such time-divisional photon counting consistently treats data output from the pixels as digital data; therefore, random noise and fixed noise associated with transmission and amplification of analog signals are not generated. At this time, only light shot noise and a dark current generated in the pixels remain, and a remarkably high S/N ratio is allowed to be obtained specifically in image pickup at low illuminance.